OpenAI release and court order

OpenAI updated ChatGPT again in April, even as a California judge ordered the company to keep one user off the service for three weeks. (help.openai.com) (reason.com) The product changes are ordinary release-note items on paper: OpenAI’s Help Center says ChatGPT in Apple CarPlay began rolling out on April 2, 2026, and mobile users got optional location sharing on March 26, 2026. The same release notes say precise location is deleted after use, but location-based answers stay in chat history unless the conversation is deleted. (help.openai.com) OpenAI’s release notes also say updated Box, Notion, Linear, and Dropbox apps started rolling out on March 27, and a simplified sidebar reached iPhone and Android users on March 26. Those entries show how often the company is changing what ChatGPT can do, where it runs, and what outside services it can touch. (help.openai.com) At the same time, a San Francisco Superior Court case turned ChatGPT access itself into the subject of an emergency order. Reason reported on April 13 that a court ordered OpenAI to cut off one user’s ChatGPT access for three weeks. (reason.com) The underlying lawsuit was filed on April 9, 2026, in San Francisco County Superior Court as case number CGC-26-635725, and the temporary restraining order application was set for an April 13 hearing before Judge Christine Van Aken. In that filing, Jane Doe asked for a temporary restraining order, an order to show cause on a preliminary injunction, and expedited discovery. (reason.com) Doe’s declaration says her ex-boyfriend became convinced in 2025 that he had invented a cure for sleep apnea with ChatGPT, then used the tool to generate false psychological reports about her and send them to friends, family, and colleagues. The declaration also says she urged him in July 2025 to seek help, and he replied that ChatGPT had rated him a “level 10 in sanity.” (reason.com) That declaration says OpenAI deactivated the user’s account in August 2025 for “Mass Casualty Weapons” activity, initially upheld the ban, and then reversed it a day later. It also says the user later showed conversations titled “Violence list expansion” and sent continued emails involving OpenAI’s Trust and Safety team. (reason.com) The temporary restraining order application says OpenAI’s lawyers told Doe’s counsel on April 9 that the company would, “pending a full review,” suspend the account and take some steps to prevent new accounts. TechCrunch reported on April 10 that Doe’s lawyers said OpenAI had agreed to suspend the user’s account but opposed other requested relief, and OpenAI did not respond in time for that report. (reason.com) (techcrunch.com) The case lands as OpenAI’s own release notes show ChatGPT spreading into cars, phones, apps, and location-aware features. In April 2026, the company was still shipping new conveniences while a court was testing whether access to the chatbot could be limited like any other tool a judge thinks poses an immediate risk. (help.openai.com) (reason.com)